Chroma subsampling is the practice of storing color information, chroma, at lower spatial resolution than the brightness channel, luma, exploiting the fact that the human eye sees brightness detail far more sharply than color. It is written as a three-part ratio: 4:4:4 keeps full color, 4:2:2 halves horizontal chroma resolution, and the common 4:2:0 halves chroma both horizontally and vertically. Counting samples makes the saving concrete: across four pixels, 4:4:4 carries twelve samples, four luma plus four Cb plus four Cr, while 4:2:0 carries six, four luma plus one Cb plus one Cr, dropping chroma from eight samples to two, a seventy-five percent cut, and halving total data. The catch is that it happens in pre-processing, before the encoder runs, so it is a quality decision teams make almost without noticing. The loss is usually invisible on faces and landscapes but shows as color bleeding on saturated text and sharp color edges, making it a major and frequently overlooked quality-loss point alongside resolution scaling.
